|Home | About | Journals | Submit | Contact Us | Français|
Integrin β1 is critical for basement membrane organization and hair follicle morphogenesis in the skin epidermis; however, less is known about its function in the developing oral epithelium. Since the skin and oral epithelia share structural similarity, we hypothesized that β1 integrin function would be critical for the normal development of oral epithelium and tooth buds. The conditional (oral mucosa-specific) β1 integrin knockout (KO) mice displayed severe disruption of the basement membrane of the tongue epithelium and developing tooth buds. Interestingly, unlike the developing hair follicles, early morphological development of the KO molar tooth buds was normal. However, subsequent morphogenetic events, such as cusp formation, cervical loop down-growth, and ameloblast polarization, did not occur normally. Primary KO oral keratinocytes showed defective cell spreading and robust focal adhesions. Our studies indicate that β1 integrin plays an essential role in the normal development of the oral epithelium and its appendages.
Integrins are a large family of heterodimeric trans-membrane receptors, comprised of αβ subunits that link the extracellular matrix (ECM) to the cellular cytoskeleton (Hynes, 2002). Integrins signal through the cell membrane in a bi-directional manner and regulate cellular processes such as cell adhesion, proliferation, migration, and basement membrane assembly (Giancotti and Ruoslahti, 1999; ffrench-Constant and Colognato, 2004; Berrier and Yamada, 2007). Numerous studies have documented that abnormal integrin expression contributes to the pathogenesis of benign and neoplastic oral tumors, underscoring the importance of these receptors in both normal and transformed cells (Breuss et al., 1995; Arihiro et al., 2000; Ahmed et al., 2002).
The predominant epithelial integrins are α3β1 and α6β4, both of which are receptors for laminin 5, a major component of the epithelial basement membrane (Watt, 2002). Basement membranes (BM) or basal laminas are specialized extracellular matrices composed of a characteristic set of large extracellular glycoproteins, arranged in an ordered fashion, which lie in direct contact with the cell surfaces (Timpl, 1996; Timpl and Brown, 1996; Yurchenco et al., 2004). The β4- and β1-containing integrins are associated with two distinct cell-substratum adhesive structures. Integrin β4, along with its partner α6, forms the central core of the hemidesmosome, which connects the ECM to the keratin cytoskeleton network within the keratinocytes (Nievers et al., 1999). The αβ1-containing integrins, in contrast, form the core of the focal adhesions, which connect the ECM to the actin cytoskeleton network with the basal keratinocytes (Raghavan et al., 2003). Loss of β4 orα6 function in the epidermis results in severe skin blistering and loss of hemidesmosome integrity (Dowling et al., 1996; Georges-Labouesse et al., 1996; van der Neut et al., 1996). The conditional knockout (KO) of β1 integrin, in contrast, results in loss of BM integrity, abnormal hair follicle morphogenesis, and defective keratinocyte migration (Brakebusch et al., 2000; Raghavan et al., 2000).
The objective of this study was to investigate the consequences of the loss of integrin β1 function in the oral epithelia. The oral epithelium shares significant structural similarity with the skin epidermis (Presland and Dale, 2000; Presland and Jurevic, 2002). Using the neonatal lethal β1 conditional KO mice, which we previously generated (Raghavan et al., 2000), we were able to explore the function of αβ1 integrins in the oral epithelium during embryonic development.
The generation of the β1 integrin floxed/K14Cre mice has been previously described (Raghavan et al., 2000). The appropriate genotype was confirmed by PCR analysis of tail DNA. All procedures were approved by the Columbia University Animal Care and Use Committees (IACUC) and were in accordance with the International Association for the Study of Pain guidelines for the care and use of laboratory animals
Staged embryos were collected by Caesarean section following CO2 narcosis. The heads were embedded in a coronal or sagittal orientation in optimal cutting temperature compound (OCT) and stored at −80°C. Frozen tissues were cut at 10-µm thickness in a Leica 1850 cryostat and then processed for histological staining or prepared for indirect immunofluorescence staining.
For morphological analysis, sections were stained with hematoxylin & eosin (H&E) according to standard procedures. Immunofluorescence analyses on tissue and keratinocytes were performed as previously described (Raghavan et al., 2000, 2003). Antibodies used were: β1, α6 integrin, collagen IV, tenascin C, HSPG (1:100; all from Millipore, Bedford, MA, USA), β6 integrin (1:250; Biogen Idec, Cambridge, MA, USA), laminin-5, laminin-α5, keratin-5/14, e-cadherin (1:100; all gift antibodies), vinculin, and paxillin (1:100, Sigma, St. Louis, MO, USA). Fluorescein isothiocyanate (FITC) and rhodamine red-X (RRX), coupled secondary antibodies (1:100; Jackson ImmunoResearch Labs, West Grove, PA, USA), were used to detect the primary antibody, and nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI). The expression of these antibodies in skin epithelia and keratinocytes was used as a positive control, and secondary antibody alone was used as a negative control. The tissue sections and cells were imaged with a Zeiss Axiophot or Zeiss 510 Meta confocal microscope (Carl Zeiss, Inc., Thornwood, NY, USA).
Oral keratinocytes were isolated from the tongue epithelia of E17.5 WT and β1 KO embryos according to a protocol previously described for isolating epidermal keratinocytes (Raghavan et al., 2003). The cells were plated on mitomycin-C-treated J2-3T3 fibroblast feeders and grown at 32°C, since this reduced the number of cells that underwent terminal differentiation due to the presence of high levels of calcium in the keratinocyte growth media. The cells were passaged after 7 to 14 days.
In the conditional β1 integrin KO mice, the expression of β1 integrin was lost from the oral epithelium of the tongue, palate, and developing tooth buds (Fig. 1B; Appendix Fig. 1B). Given that the phenotype in the skin epithelia of these mice is quite severe, we wanted to examine the status of the tongue epithelium in the β1 integrin KO embryos. Histological examination of the β1 integrin KO tongue epithelia (Fig. 1B) revealed that, as in the skin epidermis, there were large blisters at the dermal-epidermal junction (DEJ) (arrowheads; Fig. 1B). Next we looked at the status of the basement membrane using antibodies against laminin 5 (shown) and collagen IV (data not shown). As expected, in the WT tongue, the expression of laminin 5 at the DEJ appeared intact throughout the section (Fig. 1C). In contrast, the basement membrane of the KO tongue appeared to be severely disrupted, with a discontinuous expression pattern of laminin 5 at the DEJ (Fig. 1D). In the KO epithelia, de novo expression of the basement membrane protein tenascin C and β6 integrins was detected (Figs. 1F, ,1H).1H). These markers are highly up-regulated during wound healing and, interestingly, also in oral tumors (Franz et al., 2006; Thomas et al., 2006). However, increased expression of β6 integrin was unable to rescue the β1 integrin KO phenotype in the β1 integrin KO oral epithelia, suggesting that β1 and β6 integrins may have non-redundant functions. Taken together, our results suggest a critical function for β1 integrin in the tongue epithelia, including that of maintaining BM integrity. Moreover, loss of β1 integrin may elicit a wound-healing response in the oral epithelia, as judged by increased expression of tenascin C and β6 integrin.
Since the loss of β1 integrin in the skin epithelium results in loss of hair follicle morphogenesis, we were particularly interested to see if the tooth buds (which exhibit many morphological and molecular similarities with developing hair follicles) were affected in the β1 integrin conditional KO mice. Tooth morphogenesis starts at embryonic day (E) 9-11 through a series of reciprocal inductive signals between the oral ectoderm and the neural-crest-derived mesenchyme (Thesleff, 1991, 1995; Thesleff et al., 1995; Thesleff and Nieminen, 1996). These signaling events are accompanied by localized thickenings of the oral epithelium at the position of the future teeth. This thickening grows into a bud at E13.5. Next, the tip of the epithelial tooth bud folds, resulting in the formation of a cap-like structure, which surrounds the condensed mesenchyme, referred to as the dental papilla. Subsequent folding and growth of the epithelial cap eventually gives rise to the bell stage. In the β1 integrin KO embryos, the early stages of tooth bud morphogenesis—the initiation, bud, cap, and early bell stages—proceeded normally (data not shown). The first time-point when a significant difference was seen in the morphology of the KO molars was at E17.5, when the developing tooth buds were in mid-bell stage. The WT tooth buds had defined cusps, and the cervical loops extended down, encapsulating the dental papilla (Appendix Fig. 2A). In the KO tooth buds, the cusps were less-well-defined, and the cervical loops did not extend down as far (Appendix Fig. 2B). A higher-magnification image of the cells from the developing cusps revealed that, in the WT tooth bud (Appendix Fig. 2A′), the pre-ameloblasts were beginning to polarize and elongate, whereas in the KO, the pre-ameloblasts were still quite rounded and not polarized (Appendix Fig. 2B′). At E19.5 (which was the latest stage at which we could study tooth morphogenesis in the KO), the WT molars assumed the shape of the mature tooth, the cusps were very well-defined, and the cervical loops extended further down (Fig. 2A; Appendix Fig. 2C). In the KO tooth bud, in contrast, the cusps were still not well-developed, and the downward growth of the cervical loops appeared to be inhibited (Fig. 2B; Appendix Fig. 2D). Higher-magnification images of the developing WT ameloblasts (which are in a pre-secretory stage at E19.5) showed that they were well-elongated and polarized (Fig. 2C; Appendix Fig. 2C′). In the KO tooth buds, the ameloblasts were still rounded and failed to polarize (Fig. 2D; Appendix Fig. 2D′). Likewise, in the developing WT lower incisors, the ameloblasts were well-polarized and elongated, and the cervical loops were well-developed (Fig. 2E). In the β1 integrin KO incisors, the polarization of the ameloblasts and the development of the cervical loops were inhibited (Fig. 2F). Analysis of our data, taken together, suggests that while early stages of tooth morphogenesis do not require β1 integrin function, normal β1 integrin signaling is critical in later stages of tooth bud development, such as during the formation of the cusps, the down-growth of the cervical loops, and ameloblast polarization.
Since the developing KO molars displayed abnormal cusp and cervical loop morphogenesis, we were interested in the status of the basement membrane around the developing WT and KO molar buds. Previous studies have reported that loss of BM integrity around the developing tooth buds results in abnormal development of the cusps and smaller tooth buds (Fukumoto et al., 2006). The basement membrane surrounding the developing tooth buds was labeled with antibodies against collagen IV, heparin sulfate proteoglycans (HSPG) (Figs. 3A--3D),3D), laminin α5 (Appendix Figs. 1A, 1B), laminin 5, and entactin/nidogen (data not shown). In the developing WT tooth bud, the expression of collagen IV was adjacent to the inner (IE) and outer dental epithelia (OE), and the down-growing cervical loops (CL) (Fig. 3A). In the developing KO tooth buds, the expression of collagen IV was discontinuous around the inner (IE) and outer dental epithelia (OE), and was disrupted around the cervical loops (CL) that had not grown downward. The expression of HSPG was seen predominantly in the cusps of the developing WT molars (Fig. 3C), whereas in the developing KO molars, that expression was much weaker, and the cusps could not be clearly delineated (Fig. 3D). Next, we looked at the proliferation status of the developing ameloblasts by using an antibody against phospho-histone H3 (PH3). The expression of PH3 was robust in the developing WT ameloblasts (Am) as well as in the dental papilla (DP) (Fig. 3C). In the developing KO tooth bud, very few ameloblasts expressed PH3, although the expression of PH3 in the KO dental papilla was comparable with that in the WT (Fig. 3D). Higher-magnification images of the regions boxed in Figs. 3C and and3D3D (Figs. 3C′, 3D′) revealed that the number of proliferating cells in the KO epithelia was severely reduced (to approximately 25% of the WT). Analysis of our data, taken together, suggests that the loss of β1 integrin in the developing tooth buds results in the loss of BM integrity and reduced proliferative potential of the dental epithelial cells.
For better understanding of how the loss of β1 integrins might affect the oral keratinocytes, we isolated primary oral keratinocytes from E17.5 WT and KO embryos. Both the WT and KO oral keratinocytes were successfully isolated (Figs. 4A, ,4B).4B). We next investigated what effect, if any, the loss of β1 had on the ability of the KO cells to adhere to the underlying substratum. The WT and KO oral keratinocytes were plated on coverslips coated with 10 µg/mL fibronectin for 24 hrs, and the focal adhesions were labeled with antibodies against vinculin and paxillin. The β1 integrin KO cells were generally less well-spread compared with the WT cells; in addition, the actin cytoskeleton was more bundled in the KO (Fig. 4D) compared with the WT (Fig. 4C). The KO cells exhibited more robust peripheral focal adhesions (Fig. 4F), and the smaller central focal complexes found in WT keratinocytes (arrowheads, Fig. 4E) were not observed in the KO. The keratin cytoskeletal network was not affected in the KO (Fig. 4H), nor was the expression of the hemidesmosomal integrin α6β4.
Our analysis of the β1 integrin KO tongue epithelia indicates that the expression of β1 integrin is critical to the maintenance of basement membrane integrity. In addition, we showed that there was de novo expression of tenascin c and β6 integrins, two markers that are up-regulated in both wound healing and oral cancer. Interestingly, neither the up-regulation of β6 nor the expression of α6β4 in the oral epithelium was able to compensate for the loss of β1 expression, suggesting that constant signaling through the αβ1 integrins is required to maintain BM integrity.
Characterization of tooth development revealed important roles for β1 integrin in tooth bud morphogenesis. While early morphogenetic events (initiation, bud, and cap stages) occurred normally in the KO, we showed that cusps in the KO molars failed to develop, and the cervical loops failed to grown downward. The basement membrane surrounding the developing tooth was disrupted, particularly around the down-growing cervical loops, an area that perhaps received intense mechanical pressure. We also showed that the ameloblasts, or enamel-producing cells, failed to polarize and lost their proliferative potential. Analysis of these data indicates that β1 integrin function is critical for the morphogenetic movements that allow the molars to assume their final shape, and the lack of basement membrane integrity may also contribute to the defects in ameloblast proliferation and differentiation.
Finally, we showed that the primary KO oral keratinocytes were less well-spread and had more robust focal adhesions compared with the WT oral keratinocytes. In the future, we plan to focus on defining the molecular mechanisms of how and why the KO cells fail to undergo polarized migration.
We thank Sheila Violette (Stromedix, Cambridge, MA) and Paul Weinreb (Biogen Idec) for the anti-β6 antibodies, Dr. Bob Burgeson and Peter Marinkovich for the laminin 5 antibodies, and Dr. Jeff Miner for the laminin alpha 5 antibody. We also thank Faye Wang for help with sectioning the tissue samples.
Dr. S. Raghavan is supported by a Dermatology Foundation Career Award, The Department of Dermatology and College of Dental Medicine at Columbia University, and by NIH grant 1R03AR054022.
A supplemental appendix to this article is published electronically only at http://jdr.sagepub.com/supplemental.